Fire Following Earthquake Aspects of the Southern San...

Fire Following Earthquake Aspects of theSouthern San Andreas Fault Mw 7.8Earthquake Scenario

Charles R. Scawthorn,a)M.EERI

Fire following earthquake (FFE) is a significant problem in California. Poten-tial FFE were examined for the ShakeOut Scenario assuming a Mw 7.8 event ona morning in mid-November, with breezy (10 mph) low humidity conditions.FFE is a nonlinear process whose modeling does not have great precision – inmany cases the only clear result is differentiation between a few small fires ver-sus major conflagration. For the scenario, analysis indicates approximately 1,600ignitions, with the central Los Angeles basin experiencing hundreds of largefires. Estimated loss is hundreds to perhaps a thousand lives, and approximately200 million sq. ft. of residential and commercial building floor area, correspond-ing to a loss of perhaps as much as one hundred billion dollars virtually fullyinsured. Mitigation opportunities include construction of a seismically reliable re-gional saltwater pumping system to protect central Los Angeles, and automatedgas shut-off devices in densely built areas. [DOI: 10.1193/1.3574013]

INTRODUCTION

Fire following earthquake refers to series of events or stochastic process initiated by alarge earthquake. Fires following earthquakes are generally only a very significant problemin a large metropolitan area predominantly comprised of densely spaced wood buildings. Insuch circumstances, the multiple simultaneous ignitions can lead to catastrophic conflagra-tions that by far are the dominant agent of damage for that event. A large earthquake suchas a M7.8 event on the San Andreas Fault in southern California (or comparable events innorthern California, Puget Sound, or the Lower Mainland of British Columbia) combinesall the requisite factors for major conflagrations that, depending on circumstances, can be ofuniquely catastrophic proportions.

This paper was prepared as part of the ShakeOut Scenario and exercise, with the pur-pose of qualitatively describing fires following a M7.8 earthquake on the Southern SanAndreas Fault, with primary emphasis for assisting emergency planning. The scenariospecified the earthquake occurs on 13 November 2008, a day with average Novemberweather conditions, and no Santa Ana winds; that the scenario should be “realistic” and notsome “worst case”; and should address the following questions: (i) provide a realisticscenario of ignitions, fire growth and spread; (ii) How will ignitions be reported after anearthquake, how will fire departments respond, and how long will it take for the fires to beextinguished? What mutual aid agreements are in place and how will they be activated? (iii)How will damage to telecommunications, water supply, and roadway damage affectresponse? (iv) What, if any, effective mitigation actions have been undertaken elsewhere

a) SPA Risk LLC, San Francisco CA 94119

419

Earthquake Spectra, Volume 27, No. 2, pages 419–441, May 2011; VC 2011, Earthquake Engineering Research Institute

that might be practical in Southern California? (v) Briefly state the limitations of the FFEscenario, and summarize, if appropriate, research that would provide a more realistic, per-haps more challenging or detailed, scenario.

BACKGROUND

Large fires, for example measured in terms of square miles of burnt area, have not beenunique to fires following earthquakes—indeed, the great fires of London (1666) and Chi-cago (1871) are only the most noteworthy of a long succession of nonearthquake-relatedurban conflagrations. Large urban conflagrations were actually the norm in nineteenth-century America, so that long experience allowed the National Board of Fire Underwritersto state with some confidence (NBFU 1905):

“...In fact, San Francisco has violated all underwriting traditions and precedent by notburning up. That it has not done so is largely due to the vigilance of the fire department,which cannot be relied upon indefinitely to stave off the inevitable.”

While the 1906 San Francisco earthquake had major geological effects and damagedmany buildings, it was the fire that resulted in 80% of the total damage—a fire foreseen andexpected, irrespective of an earthquake. As the fire service was professionalized in the twen-tieth century, with improvements in equipment, communications, training, and organization,urban conflagrations tended largely to become a thing of the past (National Commission onFire Prevention and Control 1973). Largely, but not entirely, however, as witnessed in the1991 East Bay Hills fire, where 3,500 buildings were destroyed in a matter of hours.

Still, the two largest peace-time urban conflagrations in history have been fires follow-ing earthquakes—1906 San Francisco and 1923 Tokyo, the latter event’s fires causing thegreat majority of the 140,000 fatalities.

Much larger wildland fires also occur of course, and continue to be a major source ofloss, including in Southern California almost every year. However, historically earthquakeshave typically not caused major wildland fires.

Although a combination of a professionalized fire service, improved water supply andbetter building practices has largely eliminated non-earthquake related large urban confla-grations in the United States, there is still a gap—an Achilles heel—which is fire followingearthquake. This is due to the correlated effects of a large earthquake simultaneously caus-ing numerous ignitions, degrading building fire resistive features, dropping pressure inwater supply mains, saturating communications and transportation routes, and thus allowingsome fires to quickly grow into conflagrations that outstrip local resources. It is not suffi-ciently appreciated that the key to modern fire protection is a well-drilled rapid response byprofessional firefighters in the early stages of structural fires, arriving in time to suppress thefire while that is still relatively feasible. A typical response goal for urban fire departmentsfor example is four minutes from time of report to arrival. If suppression is delayed, due ei-ther to delayed response, or lack of water, a single structural fire can quickly spread toneighboring buildings and grow to the point where an entire municipality’s fire resourcesand perhaps even assistance from neighboring communities are required. This is for a singleignition. Simply put, most fire departments are not sized or equipped to cope with the firesfollowing a major earthquake. A major earthquake and its associated fires is a low-

CHARLES R. SCAWTHORN420

probability event for which, although it has very high potential consequences, it may not befeasible to adequately prepare. There are exceptions to this; the San Francisco Fire Depart-ment, Los Angeles City Fire Department, Vallejo Fire Department, and Vancouver (B.C.)Fire and Rescue Services have all undertaken special measures, which will be discussedbelow.

MODELING OF FIRE FOLLOWING EARTHQUAKE

The first step toward solving any problem is analyzing the problem and quantifying itseffects. A full probabilistic methodology for analysis of fire following earthquake wasdeveloped in the late 1970s (Scawthorn et al. 1981) and has been applied to major cities inwestern North America (Scawthorn and Khater 1992). A recent monograph (TCLEE 2005)and review paper (Lee et al. 2008) details the current state of the art in modeling fire follow-ing earthquake, so that only a brief review is presented here. In summary, the steps in theprocess are shown in Figure 1:

Figure 1. Fire following earthquake process (TCLEE 2005).

FIRE FOLLOWING EARTHQUAKE ASPECTS OF THE SOUTHERN SAN ANDREAS FAULT 421

Occurrence of the earthquake: Causing damage to buildings and contents, even if thedamage is as simple as knockings things (such as candles or lamps) over.

Ignition: Whether a structure has been damaged or not, ignitions will occur due to earth-quakes. The sources of ignitions are numerous, ranging from overturned heat sources, toabraded and shorted electrical wiring, to spilled chemicals having exothermic reactions, tothe friction of things rubbing together.

Discovery: At some point, the fire resulting from the ignition will be discovered, if ithas not self-extinguished (this aspect is discussed further below). In the confusion followingan earthquake, the discovery may take longer than it might otherwise.

Report: If it is not possible for the person or persons discovering a fire to immediatelyextinguish it, fire department response will be required. For the fire department to respond,a report to the fire department has to be made. Communications system dysfunction and sat-uration will delay many reports.

Response: The fire department then has to respond, but is impeded by nonfire-damageemergencies that also require response (e.g., building collapse), as well as transportationdisruptions.

Suppression: The fire department then has to suppress the fire. If the fire department issuccessful, they move on to the next incident. If the fire department is not successful, theycontinue to attempt to control the fire, but it spreads, possibly becoming conflagration. Suc-cess or failure hinges on numerous factors including water supply functionality, buildingconstruction and density, wind and humidity conditions, etc.

This process is also shown in Figure 2, which shows a fire department operations time-line. Time is of the essence for the fire following earthquake problem. In this figure, the hor-izontal axis is time, beginning at the time of the earthquake, while the vertical axis presentsa series of horizontal bars of varying width. Each of these bars depicts the development ofone fire, from ignition through growth or increasing size (size is indicated by the width ornumber of bars).

The analysis of the ShakeOut Scenario followed the above succession of eventsalthough given the resources available for the project, approximations were employed atseveral steps in the analysis. However, ignitions, response, and fire spread were modeled atthe ZIP code level, with over 700 ZIP codes considered in the analysis. We begin by brieflypresenting the scenario earthquake and associated framework, particularly its intensity dis-tribution. We then use simple rules of thumb to estimate the approximate number and distri-bution of ignitions, and compare these against resources to identify those areas where largefires may be expected to occur. We discuss citizen response and reporting, fire serviceresponse and other factors to arrive at an estimate of overall impacts. We then review oppor-tunities for mitigating the fire following earthquake problem and conclude with remarks onsome mitigation steps that might be taken.

SCENARIO EARTHQUAKE AND PREVAILING CONDITIONS

The scenario event is a M7.8 earthquake on the Southern San Andreas Fault, Figure 3.Seismological aspects are discussed by others in this issue and the MMI distribution


Figure 2. Fire department Operations Timeline (TCLEE 2005).

Figure 3. Scenario M7.8 fault trace and offset (Porter 2007).


employed here was developed by others and is shown in Figure 4 (Graves et al. 2011).Noteworthy are the high intensities MMI VIII–X along the fault (to be expected), but alsothe relatively high intensities, MMI VI–VIII throughout the Los Angeles Basin, northernOrange County and in the San Fernando Valley. For comparison, Figure 6 shows the MMIand ignition patterns for the 1994 Northridge earthquake.

The counties and populations affected by the scenario are shown in Table 1—the totalaffected population is approximately 19 million, and is distributed as shown in Figure 4 andFigure 5. Current fire service specific data for the entire region was not readily available for

Figure 4. MMI Map for M7.8 SOSAFE Scenario overlaid on population density by ZIP code.

Table 1. Counties and populations affected in the Scenario (CSAC 2007)

County2006 Estimated Total Population

(Millions)

Imperial 0.17

Kern 0.81

Los Angeles 10.3

Orange 3.1

Riverside 2.1

San Bernardino 2.0

Ventura 0.83

Total 19.31

Note that population data differs from Table 4 due to different year and data source.


this paper but based on previous data the total number of fire engines in the affected countiesis estimated at just under 2,000 (Table 2). Only fire engines (i.e., pumpers) are considered asthey apply water in urban structural fires—ladder trucks and other apparatus are also neces-sary to assist, but without fire engines, suppression of a structural fire is usually not possible.

In Southern California, November climate tends to have a “bimodal” distribution—some storms occur, with precipitation and lower temperatures, but Santa Ana conditions arerelatively prevalent, with very high winds and extremely low humidity. Indeed, the worstfire season in Southern California is October–November. For the scenario, breezy condi-tions (10 mph) and relatively low humidity were specified.

Figure 5. MMI Map for M7.8 SOSAFE Scenario overlaid on population density by ZIP code,Central Los Angeles Basin.

Table 2. Estimated number of fire engines per county

County 2006 Est. Fire Engines

Imperial 29

Kern 140

Los Angeles 586

Orange 391

Riverside 328

San Bernardino 320

Ventura 143

Total 1,937


FIRE FOLLOWING EARTHQUAKE ASPECTS

Based on methods previously developed (Scawthorn 1987, TCLEE 2005) and employ-ing population data for the region and intensity data from the scenario, the total number offire ignitions likely to occur in the scenario was calculated to be approximately 1,600, asshown in Table 3 and Figure 7. These are ignitions that require fire department response;there will be other minor ignitions that are suppressed immediately by citizens and typicallynot even reported and are not considered here.

Figure 6. 1994 Northridge ignitions and MMI overlaid on 2005 population density by ZIPcode.

Table 3. Estimated ignitions, large fires and final burnt SFED M7.8 ShakeOut Scenario(12 noon, 13 Nov 2008, 10 mph wind, low humidity)

Est No.Ignitions

Est. No.Large Fires

Est. BurntSFED (thous)

Est. Burnt Bldg.Floor Area (thous. sq. ft.)

Imperial 131 45 negligible negligible

Kern 167 82 negligible negligible

Los Angeles 612 583 94 140

Orange 206 165 37 56

Riverside 239 157 1 2

San Bernardino 234 151 1 2

Ventura 18 0 negligible negligible

Total 1,606 1,182 133 200


Table 3 shows that about 70% of the losses occur in Los Angeles County, due to about600 fires. The approximately 600 fires for Los Angeles county can be verified using for exam-ple, HAZUS-MH (FEMA 2003), which provides a regression for estimation of ignitions:

Ignitions ¼ �0:025þ ð0:592 � PGAÞ � ð0:289 � PGÞ (1)

where PGA is peak ground acceleration (g) and Ignitions is ignitions per million sq. ft. ofbuilding floor area. This regression results in 0.226 ignitions per million sq. ft. of buildingfloor area for example for PGA¼ 0.6, or on average about 1 ignition per 4.4 million sq. ft.Table 4 shows the application of this equation for Los Angeles County, which results in an

Figure 7. Ignitions (one trial) overlaid on MMI for M7.8 SOSAFE Scenario and population den-sity by ZIP code, Central Los Angeles Basin (black rectangle corresponds to ZIP code 90002).

Table 4. Alternative estimated ignitions, Los Angeles County [based on Equation 10-1 and Ta-ble 10.1 (FEMA 2003)]

MMI 2005 Population Est. Sq. Ft. (millions) PGA IGN rate Ignitions

10 258,536 136 0.71 0.250 34

9 1,420,747 746 0.6 0.226 169

8 1,625,122 853 0.45 0.183 156

7 2,148,548 1,128 0.28 0.118 133

6 2,077,710 1,091 0.16 0.062 68

<6 1,027,658 540 negligible

Total 8,558,321 560


estimate of about 560 ignitions. The HAZUS methodology was not used in this study (hencethe difference of 560 versus 612), but is cited here as corroboration for the results shown inTable 3. Comparison of the MMI patterns and population densities for the 1994 Northridgeearthquake (2005 population densities) shows why about 16 times more ignitions areexpected in the scenario event—simply put, densely populated and built-up central LosAngeles and Orange Counties are much more strongly shaken in this event than in 1994.

The cause of ignitions would likely be similar to causes in the 1994 Northridge earth-quake, which is the best U.S. data set for recent fires following an earthquake; about half ofall ignitions would be electrical-related, a quarter gas-related, and the remainder due to a va-riety of causes, including chemical reaction (Table 5). It is worth noting that, although elec-tric power often fails during the earthquake shaking in high intensity areas, electricallycaused ignitions still occur, due either to arcing before power fails, stored energy in electri-cal appliances, and/or when power is restored. Also, based on the Northridge experience,

Table 5. General sources of ignition, LAFD data, Northridge earthquake(Scawthorn et al. 1998)

Source Fraction

Electrical 56%

Gas-related 26%

Other 18%

Figure 8. Detail of Figure 7 centered about ZIP code 90002, showing ignitions and firestations.


about half of all ignitions would typically occur in single-family residential dwellings, withanother 26% in multifamily residential occupancies, that is, about 70% of all ignitions occurin residential occupancies. Educational facilities would be a small percentage of all ignitions(3% in Northridge), and most of these are due to exothermic reactions of spilled chemicalsin chemistry laboratories.

Of particular concern is the large number of oil refineries, tank farms, and related facili-ties in and around Long Beach. These facilities are responsible for half of California’s gaso-line and one third of the refined gasoline west of the Rockies. When strongly shaken, oilrefineries and tank farms have typically had large fires which have burned for days. Exam-ples include the Showa refinery in the 1964 Mw 7.5 Niigata (Japan) earthquake, the Tupracsrefinery in the 1999 Mw7.6 Marmara (Turkey) earthquake (Figure 9, Scawthorn 2000), andthe Idemitsukosan Hokkaido refinery fire in the 2003 MJMA 8.0 Tokachi-oki earthquake(Figure 10). While the Long Beach area is shown to have lower-intensity shaking, the longperiod effects at the site from the M7.8 scenario event has the potential to cause large slosh-ing in tanks and fires. To put this in perspective, the 2003 Tokachi event caused one tankfire at a 140,000 bbl/day facility 230 km from the event epicenter, while the ShakeOut Sce-nario is 80 km distant from 1.1 million bbl/day aggregate refining capacity.

The approximately 1,600 ignitions requiring fire department response will initially beresponded to by citizens. As noted, they will be able to suppress some fires, which are notactually included in the 1,600. When they realize the fire is beyond their capabilities, theywill endeavor to call the fire department, by telephone since fire alarm street pull boxeshave largely disappeared from the U.S. urban landscape. Attempts to report via 911,

Figure 9. Tupracs refinery, Mw 7.41999 Marmara (Turkey) earthquake. Photo by G. Johnsonin Scawthorn (2000).


however, will often be unsuccessful, due not so much to damage to the telephone systembut to simple saturation of the system and 911 call centers. Citizens will then travel to thenearest fire station, but such “still alarms” will be largely unneeded, since the fire companieswill have already responded to the nearest fire (“self-dispatched”), if not dispatched by 911.

Experience shows that citizens on scene will respond rationally (Van Anne 1989) rescu-ing as many people as possible and protecting exposures. Water supply from mains (dis-cussed below) will often be unavailable but in Southern California, backyard swimmingpools are a valuable and widespread resource (Scawthorn et al. 1998).

The Los Angeles City Fire Department (LAFD) and other fire departments have for severaldecades developed Community Emergency Response Teams (CERT, see http://www.cert-la.com). A total count of citizens who have undergone CERT training is not available but is severaltens of thousands. Individually, and then as organized CERT teams, these teams will save livesand make a difference. However, large conflagrations challenge even the best equipped andtrained professionals, and for these fires, the CERT teams’ contribution will be modest.

The initial response of fire companies and personnel in the region of the scenario willbe to self-protect during violent shaking, and as soon as possible, to open the doors andremove apparatus from the fire stations. Different departments have somewhat varyingearthquake procedures but in general, companies will remove apparatus to a predesignatedlocation—often simply in front of the fire station—check the station for damage, and per-form a radio check. By this time, typically within five minutes, they will either have self-dispatched to an observed smoke column, responded to a citizen still alarm, or beeninstructed to mobilize with other companies into a strike team.

Local fire service resources will be completely committed and in need of assistancefrom outside the region. The primary needs will be personnel, additional hose, hard suction

Figure 10. (a) Idemitsokan fire, 2003, in the M8 Tokachi-oki earthquake (Japan; photo credit:Fire and Disaster Management Agency of Japan).


hose, foam, light equipment (gloves, hand tools, self-contained breathing apparatus, orSCBA) and heavy equipment (cranes, bulldozers, backhoes). Additional fire apparatus(pumpers and ladder trucks) will not be the primary need, initially, but will still prove usefulas extraregional strike teams arrive.

In the initial stage, personnel needs may be significantly supplemented by CERT teamsbut will be more significantly strengthened by the recall of off-duty trained firefighters. Off-duty personnel can be expected to have doubled staffing within 3–6 hours, and tripled itwithin 12–24 hours. While responding, an issue will be how these personnel marry up withtheir companies, and there will be some inefficiencies as personnel join first available com-panies. Nevertheless, arrival of off-duty personnel will be very important, to spell on-dutypersonnel nearing their physical limits.

As noted above, 911 centers will be overwhelmed, and doing as much as possible to tri-age events and dispatch resources. Reports of fires during the initial period will be haphaz-ard. Most fire departments do not have their own helicopters, and TV helicopter newsreporting will be a valuable resource for a few major incidents, but not most. An anecdotedemonstrates this—the first knowledge the San Francisco Fire Department’s EmergencyOperations Center (EOC) had of the Marina fire in the 1989 Loma Prieta earthquake wasfrom television news reports (despite several engine companies having already responded).Quickly gaining an accurate, complete situational awareness is still a challenge.

Local, county, and state EOCs will activate within a very short period, certainly withinan hour, and in some cases more quickly. Automatic and mutual aid in the affected regionwill largely be ineffective, due to all departments having no resources to spare. The State ofCalifornia emergency services are organized into six mutual aid regions, with the scenarioearthquake occurring at the crux of three of these regions (I, V, and VI). It will take severalhours for these three regions to have a first needs assessment (longer if the earthquakeoccurs during nightfall, but this scenario assumes a 10:00 a.m. event), although the stateoffice of emergency services (OES) will already have dispatched strike teams from otherregions.

FIRE SPREAD

The initial 1,600 ignitions will not all develop into conflagrations. There are approxi-mately 2,000 fire engines in the region, and many will be close by and able to rapidlyrespond to ignitions. Figure 8 shows a detail of Figure 7, centered around ZIP code 90002in central Los Angeles, and it shows one trial simulation of ignition locations, vis-a-vis firestations. It can be seen there are more ignitions than resources so that the normal four-mi-nute structural fire response goal will hardly be met. This delayed response, due to someareas having more fires than nearby engines, as well as delayed reporting due primarily tofailure of the 911 system, will result in many of the fires on arrival having grown such thata multi-engine capacity is needed. That is, especially in low humidity conditions, anunfought ignition will grow into a room-sized fire within several minutes, and a fullyinvolved single-family structural fire within several more. To protect neighboring buildings(“exposures”), typically two or more fire companies are needed. If only one company isavailable, it’s possible that it might be able to adequately protect two exposures (using mon-itor and a hand line, with civilian assistance), but not always. In fire following earthquake


modeling, such fires, where the fire exceeds one engine company’s capabilities, are termed“large fires”. Based on these considerations, the number of large fires for the scenario eventwas calculated to be approximately 1,200 (Table 3).

This does not consider ignitions in wildland or at the wildland urban interface (WUI),which are an ongoing problem in California. Figure 13 shows, for example, the “CaliforniaFire Siege” of 2007, a series of fires in Southern California which destroyed more than3,000 homes and involved 1,400 fire engines under nonearthquake conditions.

About a third of this scenario’s large fires occur in Imperial, Kern, Riverside, and SanBernardino Counties, where building density is relatively low, so that even though the firesare initially uncontrollable, their spread within the built environment will be limited due tolow density; in effect, they will burn themselves out. Only within the more densely builtareas of Los Angeles (see Figure 15) and Orange Counties will there be relatively large firespread, developing into conflagrations. In these areas, each ignition that is not quicklyresponded to will within a few tens of minutes grow into a multiple building fire whichunder normal conditions would be a multiple alarm fire requiring the response of perhapshalf a dozen fire engines and other apparatus. Because there won’t be enough engines torespond adequately to all such fires, a significant number of fires will grow into city blockand then multiblock conflagrations. The probability of fire spread across streets and otherfire breaks was considered in this analysis and may be corroborated by data, such as thatprovided in the HAZUS fire following earthquake model, which indicates a 50% probabilityof fire spread across a street with widths typical of central Los Angeles (i.e., 100 ft. build-ing-front-to-building-front distance, as shown in Figure 15), based on light winds and noactive fire suppression. For an uncombatted fire with a 50% probability of crossing a streetdownwind and 25% probability of crossing a street upwind or sidewind, it can easily be

Figure 11. Probability of crossing firebreak (FEMA 2003).


shown that the mean total burnt area is 3.88 city blocks. Residential blocks in central LosAngeles have on average about 50 houses and low-rise commercial buildings per block,averaging about 1,500 sq. ft. per building (including outbuildings, which are numerous; seeFigure 15), so that the average burnt building floor area in light winds is about 290,000 sq.ft per uncombatted ignition. For the 583 large fires estimated to occur in Los AngelesCounty, the mean burnt area is thus about 170 million sq. ft. of total burnt building floorarea. Using more detailed calculations, actual burnt areas are somewhat less and are shownin Table 3.

The performance of lifelines, such as water supply, gas, electric power, communica-tions, and transportation, is integral to the fire following earthquake process. Others consid-ered the performance of lifelines for the ShakeOut Scenario, and this paper only briefly dis-cusses this aspect with regard to fire following earthquake. Water supply will be severelyimpacted by the scenario event. Generally, only local water supply is relevant to the fire fol-lowing earthquake process. Water pressure will drop in some portions of the more heavilyshaken area due to pipe breaks and tank failures, despite widespread efforts over the lastseveral decades to upgrade water supply systems in California. Fire departments in manyareas will have to resort to alternative water supplies (creeks, ponds, swimming pools, etc.).

Figure 12. California OES mutual aid regions.


They will be handicapped in this, since most engine companies today do not carry hard suc-tion hose, although following the Northridge earthquake, the LAFD was able to make gooduse of swimming pools using 1.5 in. siphon ejectors (Scawthorn et al. 1998). This initiallack of water supply will add to the number of large fires.

Gas-related ignitions account for about 25% of the total number of ignitions. If the numberof ignitions could be reduced from 1,600 to 1,200, the number of large fires would be decreasedin greater proportion, and the total losses further reduced. Automatic gas shut-off valves are thebest way to reduce gas-related ignitions, and should be mandated in densely built areas. TheLAFD has shown excellent leadership in seeking legislation to require gas shut-off valves.

Communications systems, particularly telephone, will sustain some damage but notenough to reduce functionality following the scenario event. However, saturation willreduce functionality to a great degree, for several hours or more. This lack of telephoneservice will result in delayed reporting, with consequences as discussed above.

As noted earlier, the scenario earthquake is at the junction of OES Regions I, V and VI,(Figure 12). Within those three regions, the only available significant fire service resources

Figure 13. Fires and wildland urban interface—California Fire Siege 2007, Day 6 (OES 2008;http://www.fire.ca.gov/fire_protection/fire_protection_2007_siege.php).


would appear to be those in the San Diego region, and OES brush rigs in the Sierra Foot-hills. It is unlikely that many resources will be made available from the San Diego region,out of concern by local governments there of a sympathetic seismic event closer to theirregion (there may also be some damage in and around San Diego, even at that distance). Amore likely source of regional resources will be a number of strike teams assembled byOES from the southern Sierra region, arriving in the affected region within 6–24 hours.While brush rigs are more suited to wildland than urban structural fires, by the time of theirarrival, the issue will be large fires that have grown into conflagrations, a situation a bitcloser to the norm for brush rigs and associated tanker trucks.

Outside the affected region, the OES is likely to stage a number of strike teams, drawngenerally from the Central Valley and the San Francisco Bay Area. One hundred striketeams, consisting of approximately 500 pumpers and other apparatus, firefighters and offi-cers, is easily within the OES’s capability, and several times this can be managed inextremis. One hundred strike teams can be assumed to arrive at staging areas within about12 hours, with probably another one hundred over the next week.

FINAL BURNT AREA

The approximately 1,200 large fires will be spread over a large area, of varying buildingdensity, and only a relatively few will grow into major conflagrations. Under the assumedwind and humidity conditions, Riverside and San Bernardino counties are each likely tosustain one or several conflagrations destroying several city blocks.

The real concern is portions of Orange County and especially the central Los Angelesbasin, where a large plain of relatively uniform dense low-rise buildings provides a fuel bedsuch that dozens to hundreds of large fires are likely to merge into dozens of conflagrationsdestroying tens of city blocks, and several of these merging into one or several super confla-grations destroying hundreds of city blocks. Two special concerns exist in this regard: (a) ifSanta Ana winds exist (which is not the assumed scenario), losses can be much larger, and(b) if extremely calm conditions exist (which is also not the assumed scenario), the potentialexists for a symmetric wind pattern to develop caused by air drawn inward by uprising airfrom super conflagrations (an example of stack effect). A self-sustaining feedback situationcan develop (commonly termed a firestorm), which can be very destructive. While relativelyunlikely, this potential should not be ignored. Concern (a) is simply a larger mass conflagra-tion, fed by higher winds. Concern (b) is potentially much worse. Both are potentiallycatastrophic.

Under the assumed scenario conditions, analysis shows that the approximately 1,200large fires will result in an ultimate burnt area equivalent to approximately 200 million sq.ft. of residential and commercial building floor area, or 133,000 single-family dwellings(SFED1). To put this in perspective, Los Angeles county (particularly central Los Angeles)

1An average California single-family dwelling is about 1,500 sq. ft. in floor area. This unit (1,500 sq. ft. floorarea) is termed a Single-Family Equivalent Dwelling (SFED) and is used to normalize and communicate overallbuilding losses in a manner readily comprehensible to lay persons. A loss of 1.5 million sq. ft. of residential andcommercial buildings for example is equivalent to 1,000 single family dwellings, or SFED. Most people canmore readily comprehend the loss of 1,000 houses, than 1.5 million sq. ft. of floor area.


will sustain about 600 fires and a total burnt area of about 140 million sq ft. of building floorarea. On average this is about 240,000 sq. ft. of building floor area burnt per fire, or about2.5 city blocks per fire, that is, loss of entire city block and loss of about three quarters ofthe blocks on either side (i.e., fire jumps one street each way, then burns out). Given thedensities of wood buildings in Los Angeles as shown in Figure 15 (discussed furtherbelow), this is not unreasonable.

IMPACTS OF FIRE FOLLOWING EARTHQUAKE SCENARIO

Estimating fatalities associated with fires following the scenario earthquake is veryproblematic. A very simple approach is taken here—in the 1991 East Bay Hills fire, whichdestroyed approximately 3,500 dwellings, 25 persons perished. The building losses pro-jected here are approximately 40 times larger. A pro rata estimate would indicate 1,000deaths due to fire following earthquake, but such an approach is admittedly very simplistic.However, hundreds of deaths directly attributable to fire following earthquake is a conserva-tively low estimate. Injuries would probably be an order of magnitude greater. Shelter needsdirectly attributable to fire following earthquake are estimated to be hundreds of thousandsof persons.

The ultimate burnt area of approximately 200 million sq. ft. of building floor area equa-tes to approximately $40 billion of building value.2 Value of contents and other improve-ments (e.g., landscaping), will only increase this loss. An additional loss is loss of use – thatis, the persons normally living in these destroyed buildings (or conducting business inthem) must find other accommodations, which will most likely not be available in the LosAngeles basin given the scenario event. This loss, termed additional living expenses by theinsurance industry (for residential occupancies) and business interruption (for commercialoccupancies), can be quite significant, equivalent to many tens of billions of dollars.Accounting for this is problematic—if persons who have lost their dwellings are housed in

Figure 14. (a) Los Angeles County drainage and storm drain system, and (b) photo of largerdrainage channel.

2Based on replacement cost of $200 per square foot; note that this is a conservatively low estimate of replacementcost at current (2008) prices.


a hotel at insurance company expense, the accounting is easy; it’s the hotel bill. But if theyare forced to live in tents following the event, at public expense, there may be no bill.3 Insuch a case, the persons haven’t paid for their tent and can’t therefore claim against the in-surance company for a financial loss. However, they have lost value in services (of theirhouse) approximately equivalent to the rental value of their house (minus the rental value ofthe tent) but won’t be compensated for those losses. While something of an accountingexercise, this is a loss that should be accounted for, overall.

Since virtually all buildings and contents in the United States are insured for fire, andU.S. insurance contracts include fire following earthquake losses under the fire policy, thedirect fire following earthquake losses for the scenario event could result in a loss approach-ing $100 billion of insurance claims. Losses of this magnitude are probably sustainable bythe U.S. insurance industry, with some strains. For comparison, the 1991 East Bay Hills fireresulted in the loss of 3,500 homes with about $1 billion in insurance payments (1991 dol-lars). The event projected here is 17 years of inflation later and about 40 times as large. Thefire following earthquake losses are likely to be the largest portion of the insured losses inthe scenario event, and could result in major distortions within the industry.

Another aspect of the economic impacts is the loss of tax revenues. A loss of $50 billionin value of improvements is likely to result in a decrease in regional real estate tax revenues

Figure 15. (a) and (b) show an area centered above the 110–105 freeway intersection, approxi-mately equivalent to 200 million sq. ft. of building floor area. (c) shows three photos that zoomin on the area just northeast of the 110–105 freeway intersection, to show a high density ofwood buildings, typical of much of the Los Angeles Basin. Note typical 100 ft. street width(measured building-front-to-building-front).

3However, public authorities may attempt to recoup their expenses from insurance companies if the sheltered per-sons are insured.


approaching a billion dollars, for several years, directly attributable to fire followingearthquake.

MITIGATION OF FIRE FOLLOWING EARTHQUAKE

Mitigation of fire following earthquake has been extensively discussed elsewhere (TCLEE2005) so that only some limited observations specific to the scenario are provided here.

FIRE SERVICE OPPORTUNITIES

The fire service in Southern California is among the finest in the world and perhaps thebest practiced in the world in dealing with large conflagrations, due to the wildland firesrecurring annually in the region. The fire service has also been relatively diligent in prepar-ing for a large earthquake; the CERT program is a model in that regard. However, the fol-lowing mitigation opportunities are cited below, to name a few:

• Capability for more quickly assessing the incident and facilitating incident report-ing, should be improved. Reconnaissance using unmanned aerial vehicles (UAVs),and mobile phone text messaging incident reports directly to a 911 portal, shouldbe developed and operationalized. In this regard, mobile phone service will be cru-cial, but to the best of this author’s knowledge, cellular phone towers currently donot have backup power.

• Alternative water supply capability needs to be enhanced. Hard suction hosesshould be carried on all engines. Large diameter hose (LDH) systems, comparableto the San Francisco Fire Department’s Portable Water Supply System (PWSS;Scawthorn et al. 2006), should be developed on a regional basis.

• Los Angeles currently has little ability to access seawater and move it significantdistances inland—relaying via street-laid hose and engines is not an efficient wayof doing this. A special saltwater pumping system, similar to that of San Francis-co’s (built following the 1906 event) and Vancouver’s (built in the 1990s and based

Table 6. Property use for 77 LAFD earthquake-related fires, 4:31 to 24:00hrs, 17 January 1994 (Scawthorn et al. 1998)

General Property Use Fraction

One or Two Family Residential 45%

Multi-Family Residential 26%

Public Roadway 8%

Office 5%

Primary/Secondary School 3%

Vacant Property 3%

Restaurant 1%

Commercial 1%

Power Production/Distribution 1%

Other 5%

Unknown 1%


on observations in the 1989 Loma Prieta earthquake) is quite feasible for LosAngeles. Several saltwater pumping stations could be built (e.g., Santa Monica,LAX airport, Los Angeles Harbor) and large-diameter seismically resistant pipecould be laid in the Los Angeles River and other river channels and the County’sextensive storm drain system (see Figure 14) to form a looped high-pressure sys-tem, accessible from high-pressure hydrants.

• A regional task force should be formed within the fire service, to examine urbanconflagration potential in more detail. The task force should be multidisciplinary.

WATER SERVICE OPPORTUNITIES

The water service in Southern California has done a lot to prepare for a major earth-quake, but more can of course still be done. One overriding issue with regard to fire follow-ing earthquake is that water agencies typically aren’t institutionally responsible for fireprotection. That is, while they provide hydrants, if the hydrants fail to supply water, theyliterally (legally) aren’t responsible. This is not to say they don’t care, but simply that thereis an institutional gap, which tends to result in water system seismic upgrades more typicallyoriented to maintenance of customer service and to minimizing direct damage to the system,rather than to maximizing firewater supply reliability. A mandate needs to be developed tomake water agencies more responsive to this need. Given the realities of water in California,this may be unlikely to occur, but should at least be considered. A real way in which wateragencies could be more responsive to the fire following earthquake problem is if each agencywere to configure and upgrade their system such that they provide a “backbone” system ofwater mains of high seismic reliability, that provided water to major sections of the commu-nity and from which the fire service could draw water to feed water to a conflagration viaLDH systems (see also the discussion above regarding a saltwater looped system).

ENERGY INDUSTRY OPPORTUNITIES

The gas utility industry could contribute significantly to reducing the fire following earth-quake problem by developing a program to either install automated gas shut-off valves or rede-signed meters with seismic shutoffs in densely built up areas. More problematic are opportunitiesin regard to electricity. Electric power often fails in large earthquakes, due to automatic systemtrips as well as damage to the system. However, the power failure usually takes several seconds,during which power is a source of many ignitions. Certain electric appliances (e.g., those withheating elements) can still cause fires even after power is cut. Large-scale intentional curtailmentof power is problematic, since some communications and other essential equipment would thenbe useless. The petroleum refineries and related facilities in the Long Beach area are likely to sus-tain major fires in the scenario event. Their earthquake preparedness should be reviewed.

CONCLUDING REMARKS

Fires follow all earthquakes affecting human settlements but are potentially catastrophic phe-nomena in selected areas, such as Southern California. A large earthquake will occur on theSouthern San Andreas Fault, perhaps similar to the scenario considered here. Just as the fires fol-lowing the 1906 earthquake were quite foreseeable, the fires following a ShakeOut-like event areforeseeable and will likely constitute a significant portion of the overall impacts of that event.


To put the loss estimates presented in this paper in perspective, Figure 15 is a series ofaerial photos of the Los Angeles Basin taken from Google Earth that make two key points:

• The estimated 200 million sq. ft. of burnt building floor area, while an enormousloss, is only a small fraction of the exposure (1.5%). The red rectangles indicate theequivalent area (very approximately) relative to the total exposure.

• The images show the high density of wood buildings typical of the central LosAngeles Basin. Note the small interbuilding spacing and almost total building cov-erage of many blocks. While there are broader avenues and even freeways whichserve as firebreaks, flying brands can easily drift thousands of feet and/or severalmiles downwind, easily crossing such firebreaks (this was seen, for example, in the1991 East Bay Hills fire, where the fire jumped Highway 24, a ten-lane freeway).

As part of the ShakeOut Scenario development, the findings of this paper were reviewed by apanel of representatives of major fire departments in California, including Chief Donald Manning(Los Angeles City Fire Department, retired and former Chair, California Seismic Safety Commis-sion), Chief Don Parker (Vallejo Fire Department, retired and Chair, California Seismic SafetyCommission) and Captain Larry Collins (Los Angeles County Fire Department), all of whom con-curred with its findings, with the only demur being that the estimates “might be a bit low.”

While foreseeable, quantification of the fire following earthquake risk is still very impre-cise. The only previous quantified estimates of fire following earthquake risk for SouthernCalifornia were made one to two decades ago (Scawthorn and Khateer 1992, Scawthorn1987), and this paper is only a very approximate estimate. The size and importance of theproblem warrants much more detailed analysis using the latest data and methods.

REFERENCES

California State Association of Counties (CSAC), 2007. Organization website, http://www.csac.counties.org/default.asp?id¼399, accessed 28 December 2007.

Federal Emergency Management Agency (FEMA), 2003. Multihazard Loss Estimation Method-ology, Earthquake Model, HAZUS-MH MR3 Technical Manual, developed by the Depart-ment of Homeland Security, Emergency Preparedness and Response Directorate, FEMAMitigation Division Washington, D.C., under a contract with the National Institute of Build-ing Sciences, Washington, D.C., 699 pp.

Graves, R. W., Aagaard, B. T., and Hudnut, K. W., 2011. The Shake Out earthquake source andground motion simulations, Earthquake Spectra 27, 273–291.

Lee, S., Davidson, R., Ohnishi, N., and Scawthorn, C., 2008. Fire following earthquake—Reviewing the state-of-the-art of modeling, Earthquake Spectra 24, 933–967.

National Commission on Fire Prevention and Control, 1973. America Burning: The Report ofthe National Commission on Fire Prevention and Control, report to the President of theUnited States of America, 191 pp.

National Board of Fire Underwriters (NBFU), 1905. Report of National Board of FireUnderwriters by its Committee of Twenty on the City of San Francisco, Henry Evans,Chmn., 64 pp.

Office of Emergency Services (OES), 2008. California Fire Siege 2007—An Overview, Califor-nia Governor’s Office of Emergency Services, together with California Department of For-estry and Fire Protection and the U.S. Forestry Service, Sacramento, CA, 110 pp.


http://www.csac.counties.org/default.asp&hx003F;id&hx003D;399



http://dx.doi.org/10.1193/1.2977493

Scawthorn, C., 1987. Fire following earthquake: Estimates of the conflagration risk to insuredproperty in greater Los Angeles and San Francisco, All-Industry Research Advisory Council,Oak Brook, IL.

Scawthorn, C., 2000. The Marmara, Turkey Earthquake of August 17, 1999: ReconnaissanceReport. MCEER Tech. Rpt. MCEER-00-0001, Multidisciplinary Center for Earthquake Engi-neering Research, SUNY, Buffalo.

Scawthorn, C., Cowell, A. D., and Borden, F., 1998. Fire-Related Aspects of the NorthridgeEarthquake, NIST GCR 98-743, National Institute of Standards and Technology Buildingand Fire Research Laboratory, Gaithersburg, MD, 165 pp.

Scawthorn, C., and Khater, M., 1992. Fire Following Earthquake: Conflagration Potential inthe Greater Los Angeles, San Francisco, Seattle, and Memphis Areas, EQE International,prepared for the National Disaster Coalition, San Francisco.

Scawthorn, C., O’Rourke, T. D., and Blackburn, F. T., 2006. The 1906 San Francisco earth-quake and fire—enduring lessons for fire protection and water supply, Earthquake Spectra22, S135–S158.

Scawthorn, C., Yamada, Y., and Iemura, H., 1981. A model for urban postearthquake fire haz-ard. Disasters 5, 125–132.

Technical Council on Lifeline Earthquake Engineering (TCLEE), 2005. Fire Following Earth-quake, Scawthorn, C., J. M. Eidinger, A. J. Schiff (Editors), Technical Council on LifelineEarthquake Engineering Monograph No. 26, American Society of Civil Engineers, Reston,VA, 345 pp.

Van Anne, C., Scawthorn, C., eds., 1989. The Loma Prieta California Earthquake of October17, 1989—Fire, Police, Transportation, and Hazardous Materials: Societal Response, U.S.Geological Survey Professional Paper 1553-C. U.S. Government Printing Office, Washing-ton, D.C., 1994, D. S. Mileti, coordinator, 44 pp.

(Received 8 January 2010; accepted 10 January 2011)


http://dx.doi.org/10.1193/1.2186678

http://dx.doi.org/10.1111/j.1467-7717.1981.tb01095.x

Fire Following Earthquake Aspects of the Southern San...

Documents

Transcript of Fire Following Earthquake Aspects of the Southern San...